Samuel Tardieu

> `u16_value < b'a'` That won't compile, types are incompatible.

@sandersaares If you could rebase your PR on the latest master, this would help to have a clean CI build.

I was rather against using this possibility when Fedilab introduced it in its Android application, but it has since uncluttered my timeline (many of my contacts use Fedilab or another...

I have separated the style fixes into various commits so that some could be removed if the maintainers prefer to let the code this way. This can be the case...

Rebased on current master branch.

Hi Josh, I'm not sure the current API can readily support a bidirectional search. In addition to the points you've already mentioned (the `successors` API not being inherently reversible in...

Hi. You are right, the current edges representation does not allow edges distinction. The kind of graph that can be represented should probably be described in the documentation, or we...

I wonder if this is the right fix. GNU `mv` fails at moving `t/a` into `u/b` if `u/b` exist and is the same file as `t/a` (hard link), while your...

I think you can reuse/move around most of the code just before your patch. IIRC, it checks for hardlink equivalence when the target is a file, so if you build...

Your PR compiles on my Android 64bit system. The result of the test is: ``` failures: ---- test_df::test_type_option_with_file stdout ---- run: /data/data/com.termux/files/home/Dev/coreutils/target/debug/coreutils df --output=fstype . run: /data/data/com.termux/files/home/Dev/coreutils/target/debug/coreutils df -t f2fs...